No abstract

Purpose This prospective clinical study assesses the feasibility of training a deep neural network (DNN) for intravoxel incoherent motion (IVIM) model fitting to diffusion‐weighted MRI (DW‐MRI) data and evaluates its performance. Methods In May 2011, 10 male volunteers (age range, 29–53 years; mean, 37) underwent DW‐MRI of the upper abdomen on 1.5T and 3.0T MR scanners. Regions of interest in the left and right liver lobe, pancreas, spleen, renal cortex, and renal medulla were delineated independently by 2 readers. DNNs were trained for IVIM model fitting using these data; results were compared to least‐squares and Bayesian approaches to IVIM fitting. Intraclass correlation coefficients (ICCs) were used to assess consistency of measurements between readers. Intersubject variability was evaluated using coefficients of variation (CVs). The fitting error was calculated based on simulated data, and the average fitting time of each method was recorded. Results DNNs were trained successfully for IVIM parameter estimation. This approach was associated with high consistency between the 2 readers (ICCs between 50% and 97%), low intersubject variability of estimated parameter values (CVs between 9.2 and 28.4), and the lowest error when compared with least‐squares and Bayesian approaches. Fitting by DNNs was several orders of magnitude quicker than the other methods, but the networks may need to be retrained for different acquisition protocols or imaged anatomical regions. Conclusion DNNs are recommended for accurate and robust IVIM model fitting to DW‐MRI data. Suitable software is available for download.

This is an open access article under the terms of the Creat ive Commo ns Attri bution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

Background Palliative systemic treatment in patients with advanced or metastatic esophagogastric cancer may result in improved overall survival and quality of life but can also lead to considerable toxicity. In various cancer types, severe muscle mass depletion (sarcopenia) and poor muscle strength are associated with decreased survival and increased chemotherapy‐related toxicity. The aim of this study is to determine the impact of body composition on survival and chemotherapy toxicity in esophagogastric cancer patients treated with first‐line palliative chemotherapy. Methods A total of 88 patients with advanced esophagogastric cancer treated with standard first‐line palliative systemic therapy consisting of capecitabine and oxaliplatin (CapOx) between January 2010 and February 2017 were included. Skeletal muscle index (SMI), reflecting muscle mass, and skeletal muscle density (SMD), associated with muscle strength, were measured using pre‐treatment of all patients and evaluation computed tomography scans after three treatment cycles of 65 patients and were used to determine sarcopenia and sarcopenic obesity (i.e. sarcopenia and body mass index >25 kg/m 2 ). The associations between body composition (SMI, SMD, sarcopenia, and sarcopenic obesity) and survival and toxicity were assessed using univariable and multivariable Cox and logistic regression analyses, respectively. Results Of 88 patients, 75% was male, and median age was 63 (interquartile range 56–69) years. The majority of patients had an adenocarcinoma (83%). Before start of treatment, 49% of the patients were sarcopenic, and 20% had sarcopenic obesity. Low SMD was observed in 50% of patients. During three cycles CapOx, SMI significantly decreased, with a median decrease of 4% (interquartile range −8.6–−0.4). Median progression‐free and overall survival were 6.9 and 10.1 months. SMI, SMD, sarcopenia, and sarcopenic obesity (both pre‐treatment and after three cycles) were neither associated with progression‐free nor overall survival. Pre‐treatment SMD was independently associated with grade 3–4 toxicity (odds ratio 0.94; 95% confidence interval 0.89–1.00) and sarcopenic obesity with grade 2–4 neuropathy (odds ratio 3.82; 95% confidence interval 1.20–12.18). Conclusions Sarcopenia was not associated with survival or treatment‐related toxicity in advanced esophagogastric cancer patients treated with CapOx. Pre‐treatment sarcopenic obesity was independently associated with the occurrence of grade 2–4 neurotoxicity and skeletal muscle density with grade 3–4 toxicity.

A RF model predicting 3-year overall survival based on pretreatment CT radiomic features was developed and validated in two independent datasets of esophageal cancer patients. The radiomics model had better prognostic power compared to the model using standard clinical variables.

The intravoxel incoherent motion (IVIM) model for diffusion-weighted imaging (DWI) MRI data bears much promise as a tool for visualizing tumours and monitoring treatment response. To improve the currently poor precision of IVIM, several fit algorithms have been suggested. In this work, we compared the performance of two Bayesian IVIM fit algorithms and four other IVIM fit algorithms for pancreatic cancer imaging. DWI data were acquired in 14 pancreatic cancer patients during two MRI examinations. Three different measures of performance of the fitting algorithms were assessed: (i) uniqueness of fit parameters (Spearman’s rho); (ii) precision (within-subject coefficient of variation, wCV); and (iii) contrast between tumour and normal-appearing pancreatic tissue. For the diffusivity D and perfusion fraction f, a Bayesian fit (IVIM-Bayesian-lin) offered the best trade-off between tumour contrast and precision. With the exception for IVIM-Bayesian-lin, all algorithms resulted in a very poor precision of the pseudo-diffusion coefficient D* with a wCV of more than 50%. The pseudo-diffusion coefficient D* of the Bayesian approaches were, however, significantly correlated with D and f. Therefore, the added value of fitting D* was considered limited in pancreatic cancer patients. The easier implemented least squares fit with fixed D* (IVIM-fixed) performed similar to IVIM-Bayesian-lin for f and D. In conclusion, the best performing IVIM fit algorithm was IVM-Bayesian-lin, but an easier to implement least squares fit with fixed D* performs similarly in pancreatic cancer patients.

Changes in T2 (∗) and ΔB0 are sequence-independent measures for potential visibility and artifact size, respectively. Improved visibility of markers correlates strongly to signal shift artifacts; therefore, marker choice will depend on the clinical purpose. When visibility of the markers is most important, markers that contain iron are optimal, preferably in a folded configuration. For artifact sensitive imaging, small ironless markers are best, preferably in a stretched configuration.

In this study we investigate a CT radiomics approach to predict response to chemotherapy of individual liver metastases in patients with esophagogastric cancer (EGC). In eighteen patients with metastatic EGC treated with chemotherapy, all liver metastases were manually delineated in 3D on the pre-treatment and evaluation CT. From the pre-treatment CT scans 370 radiomics features were extracted per lesion. Random forest (RF) models were generated to discriminate partial responding (PR, >65% volume decrease, including 100% volume decrease), and complete remission (CR, only 100% volume decrease) lesions from other lesions. RF-models were build using a leave one out strategy where all lesions of a single patient were removed from the dataset and used as validation set for a model trained on the lesions of the remaining patients. This process was repeated for all patients, resulting in 18 trained models and one validation set for both the PR and CR datasets. Model performance was evaluated by receiver operating characteristics with corresponding area under the curve (AUC). In total 196 liver metastases were delineated on the pre-treatment CT, of which 99 (51%) lesions showed a decrease in size of more than 65% (PR). From the PR set a total of 47 (47% of RL, 24% of initial) lesions were no longer detected in CT scan 2 (CR). The RF-model for PR lesions showed an average training AUC of 0.79 (range: 0.74–0.83) and 0.65 (95% ci: 0.57–0.73) for the combined validation set. The RF-model for CR lesions had an average training AUC of 0.87 (range: 0.83–0.90) and 0.79 (95% ci 0.72–0.87) for the validation set. Our findings show that individual response of liver metastases varies greatly within and between patients. A CT radiomics approach shows potential in discriminating responding from non-responding liver metastases based on the pre-treatment CT scan, although further validation in an independent patient cohort is needed to validate these findings.